Autism spectrum disorder (ASD) impacts almost 2% of children born today, yet very little is known regarding how to positively alter the outcomes of affected children. On the one hand, many, including the American Academy of Pediatrics, believe that early universal screening at well baby check?ups is an important step towards a positive outcome because it can lead to early treatment. In contrast, the United States Preventive Services Task Force (USPSTF) recently failed to endorse early universal ASD screening, noting that the benefits of doing so are poorly understood. What is needed to inform the debate is to examine the outcomes of a large cohort of children detected very early via universal screening at well?baby checks, and to compare them to children who did not participate in an early detection program. Here we propose to do just that: We will examine the school age outcomes (age 6?10 years) of an unprecedented sample size of 242 children with ASD detected very early in San Diego and Phoenix (i.e., ?west coast cohort?) through our Get SET Early program, which involved screening with the CSBS at well baby check? ups (mean age 17.7 months, range 12?24 months), and immediate referral for comprehensive evaluation and treatment if warranted. They will then be compared to a cohort of 242 ASD community children matched on age, gender, race, ethnicity, and SES who did not participate in our early detection program (Total N = 484). Given the rarity and uniqueness of our cohort, we plan to characterize outcomes not only on traditional measures of cognition, social behavior, etc., but also on outcome on school achievement (e.g., fully mainstreamed and/or lost their diagnosis) and family functioning (do families experience less stress?). Since the national mean age of ASD diagnosis is around 4 years, we expect that children in the community cohort will have later ages of diagnosis and poorer outcomes than those identified early via the Get SET Early program. With scientific rigor and reproducibility in mind, we will proactively test the replication of findings in an independent cohort (N=103) of toddlers screened in Boston, Philadelphia, and New Haven (i.e., ?east coast cohort?) through Project Early and a matched community cohort from the same region (Total N=120). Our specific aims are: AIM 1a: we will identify clinically meaningful outcome subtypes of ASD in our west coast cohort using unbiased network clustering approaches. AIM 1b: with this unique longitudinal cohort, we will examine changes in symptom profile, IQ and adaptive functioning between toddler and school ages. AIM 1c: we will evaluate program impact by comparing the outcomes children in our early?detected to the community cohort. AIM 1d: we will examine how well findings are replicated in our East coast sample. AIM 2: using complimentary regression and machine learning techniques, with our total sample collapsed across both west and east coast cohorts (N=602), we will test our model that earlier age at identification and treatment leads to improved outcomes. To examine other factors relating to a good outcome, moderating variables such as SES and level of treatment participation will also be included in our models. AIM 3: since state context (e.g., policies, guidelines) could also play a role in outcomes, in we plan to collect key state?level information to place our findings in context.